Marketing verbal skill generation method and system, computer equipment and storage medium

Abstract

The invention discloses a marketing verbal skill generation method and system, computer equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: collecting elements needed for generating the marketing verbal skill of a product as input data; based on the input data, generating a corresponding specific marketing verbal skill as output data through a preset large language model; based on the input data and the output data, generating a corresponding marketing verbal skill generation model by performing fine tuning on a preset pre-training model; and generating a target marketing verbal skill required by a target product based on the marketing verbal skill generation model. According to the method provided by the invention, the marketing verbal skill generation model in the vertical field can be obtained through training to generate the personalized target marketing verbal skill required by the target product, so that the marketing operation cost of the target product can be reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a marketing speech generation method, system, computer device and non-volatile computer-readable storage medium. Background Art

[0002] At present, with the rapid development of artificial intelligence technology, generative artificial intelligence (AIGC) has made significant progress in the field of text generation. Since the release of large-scale general language models such as ChatGPT, their powerful ability in text generation has attracted widespread attention. These models can generate coherent and natural text content by learning a large amount of text data, and even show the ability to assist or replace human work in some cases.

[0003] However, despite the excellent performance of general language models in text generation, they still have some limitations, especially in vertical applications. The knowledge of these models has a deadline, that is, their knowledge base is usually only updated to a certain point in time and cannot reflect the latest industry dynamics. In addition, due to the lack of in-depth understanding of specific vertical fields, these models may not be accurate or have limitations when generating professional words in specific fields.

[0004] Currently, marketing script generation tools on the market, such as applications developed based on models such as ChatGPT or ChatGLM, mostly use input-output prompting to generate text. Although this method is end-to-end and easy to implement, it has the following problems:

[0005] 1. Prompt optimization is difficult: In order to obtain better generation results, it is necessary to constantly try and optimize the prompt, which is a time-consuming process that requires professional knowledge;

[0006] 2. Insufficient vertical domain knowledge: The generated marketing words may lack the depth of knowledge in a specific industry, resulting in the words being relatively fixed and general, and lacking in specificity;

[0007] 3. Data privacy and security issues: Especially in vertical fields within the enterprise, due to data sensitivity, it is impossible to directly use external commercial large models, and customized solutions are required.

[0008] For example, in industries such as insurance, the writing of insurance product marketing scripts is usually done manually by product operators, which leads to the following problems:

[0009] 1. High operating costs: Manually writing scripts requires a lot of human resources, which increases operating costs;

[0010] 2. Polarization phenomenon: There are abundant marketing resources for popular products, but there is a serious shortage of marketing resources for long-tail products or niche products;

[0011] 3. Lack of personalization: The existing sales pitch fails to effectively combine the characteristics of the marketing population and highlight the highlights and responsibilities of the product.

[0012] To sum up, how to provide a marketing script generation method, system, computer device and non-volatile computer-readable storage medium, which can obtain a marketing script generation model in a vertical field through training to generate personalized target marketing scripts required for target products, is a problem that needs to be urgently solved by technical personnel in this field. Summary of the invention

[0013] In view of the above-mentioned deficiencies in the prior art, the object of the present invention is to provide a marketing script generation method, system, computer device and non-volatile computer-readable storage medium that can be used in financial technology or other related fields, aiming to solve the problem of how to obtain a marketing script generation model in a vertical field through training to generate personalized target marketing scripts required for target products.

[0014] In order to achieve the above object, the present invention adopts the following technical solutions:

[0015] A method for generating marketing words, comprising:

[0016] Collect the elements needed to generate product marketing pitches as input data;

[0017] Based on the input data, a preset large language model is used to generate corresponding specific marketing words as output data;

[0018] Based on the input data and the output data, a corresponding marketing speech generation model is generated by fine-tuning a preset pre-trained model;

[0019] Based on the marketing script generation model, target marketing scripts required for the target product are generated.

[0020] In a further technical solution, the marketing script generation method, wherein the elements required for collecting and generating the marketing script of the product as input data, includes:

[0021] Collect existing marketing scripts for several products, analyze and disassemble the existing marketing scripts, obtain the elements required to generate the marketing scripts for the corresponding products, and use the elements as input data; wherein the elements include: the theme, the key points of the theme's scripts, product information, and the marketing group.

[0022] In a further technical solution, the marketing speech generation method, wherein the step of generating corresponding specific marketing speech as output data based on the input data through a preset large language model, comprises:

[0023] Inputting the input data into a preset large language model to generate corresponding initial marketing words;

[0024] After data enhancement processing is performed on the initial marketing words, corresponding specific marketing words are obtained, and the specific marketing words are used as output data.

[0025] In a further technical solution, the marketing speech generation method, wherein the corresponding marketing speech generation model is generated by fine-tuning a preset pre-trained model based on the input data and the output data, comprises:

[0026] Based on the input data and the output data, the Qwen-14B model is fine-tuned by using the loRA method to generate a corresponding initial model;

[0027] The initial model is verified and tested, and when it is detected that the verification result and the test result meet the preset requirements, a corresponding marketing speech generation model is generated.

[0028] In a further technical solution, the marketing speech generation method, wherein the step of generating the target marketing speech required for the target product based on the marketing speech generation model, comprises:

[0029] Obtain product information corresponding to the target product given by the target user;

[0030] Based on the corresponding product information, determine the target theme and target marketing points of the target product, as well as the target marketing population;

[0031] The corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product.

[0032] In a further technical solution, the marketing speech generation method, wherein the corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product, comprises:

[0033] The target marketing words are sorted out and returned to the target user.

[0034] In a further technical solution, the marketing script generation method, wherein the elements include: theme, key points of the theme's script, product information, and marketing population, wherein the product information includes: product name, product highlights, and guarantee responsibility.

[0035] A marketing speech generation system, comprising:

[0036] A collection module is used to collect the elements required to generate the marketing script of the product as input data;

[0037] A first generating module is used to generate corresponding specific marketing words as output data through a preset large language model based on the input data;

[0038] A fine-tuning module, configured to generate a corresponding marketing speech generation model by fine-tuning a preset pre-training model based on the input data and the output data;

[0039] The second generation module is used to generate target marketing words required for the target product based on the marketing words generation model.

[0040] A computer device, wherein the computer device comprises at least one processor; and

[0041] a memory communicatively connected to the at least one processor; wherein,

[0042] The memory stores a computer program that can be executed by the at least one processor. When the computer program is executed by the at least one processor, the marketing speech generation method as described in any one of the above items can be implemented.

[0043] A non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, it can implement the marketing speech generation method as described in any one of the above items.

[0044] Compared with the prior art, the present invention provides a method, system, computer device and non-volatile computer-readable storage medium for generating marketing words, wherein the method includes: collecting the elements required to generate the marketing words of the product as input data; based on the input data, generating the corresponding specific marketing words as output data through a preset large language model; based on the input data and the output data, fine-tuning the preset pre-trained model to generate a corresponding marketing words generation model; based on the marketing words generation model, generating the target marketing words required for the target product. In this way, the method of the present invention can obtain a marketing words generation model in a vertical field through training to generate personalized target marketing words required for the target product, thereby reducing the marketing operation cost of the target product. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A flowchart of a method for generating marketing speech provided in an embodiment of the present invention.

[0047] Figure 2 A schematic diagram of the functional modules of a marketing speech generation system provided in an embodiment of the present invention.

[0048] Figure 3 A schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0050] In the description of the present invention, the terms "include", "including", "have", "contain", etc. are all open terms, which mean including but not limited to. The descriptions with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", etc. mean that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. The order of steps involved in each embodiment is used to schematically illustrate the implementation of the present application, and the order of steps therein is not limited and can be appropriately adjusted as needed.

[0051] Various non-limiting embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] At present, with the rapid development of artificial intelligence technology, generative artificial intelligence (AIGC) has made significant progress in the field of text generation. Since the release of large-scale general language models such as ChatGPT, their powerful ability in text generation has attracted widespread attention. These models can generate coherent and natural text content by learning a large amount of text data, and even show the ability to assist or replace human work in some cases.

[0053] However, despite the excellent performance of general language models in text generation, they still have some limitations, especially in vertical applications. The knowledge of these models has a deadline, that is, their knowledge base is usually only updated to a certain point in time and cannot reflect the latest industry dynamics. In addition, due to the lack of in-depth understanding of specific vertical fields, these models may not be accurate or have limitations when generating professional words in specific fields.

[0054] Currently, marketing script generation tools on the market, such as applications developed based on models such as ChatGPT or ChatGLM, mostly use input-output prompting to generate text. Although this method is end-to-end and easy to implement, it has the following problems:

[0055] 1. Prompt optimization is difficult: In order to obtain better generation results, it is necessary to constantly try and optimize the prompt, which is a time-consuming process that requires professional knowledge;

[0056] 2. Insufficient vertical domain knowledge: The generated marketing words may lack the depth of knowledge in a specific industry, resulting in the words being relatively fixed and general, and lacking in specificity;

[0057] 3. Data privacy and security issues: Especially in vertical fields within the enterprise, due to data sensitivity, it is impossible to directly use external commercial large models, and customized solutions are required.

[0058] For example, in industries such as insurance, the writing of insurance product marketing scripts is usually done manually by product operators, which leads to the following problems:

[0059] 1. High operating costs: Manually writing scripts requires a lot of human resources, which increases operating costs;

[0060] 2. Polarization phenomenon: There are abundant marketing resources for popular products, but there is a serious shortage of marketing resources for long-tail products or niche products;

[0061] 3. Lack of personalization: The existing sales pitch fails to effectively combine the characteristics of the marketing population and highlight the highlights and responsibilities of the product.

[0062] To sum up, how to provide a marketing script generation method, system, computer device and non-volatile computer-readable storage medium, which can obtain a marketing script generation model in a vertical field through training to generate personalized target marketing scripts required for target products, is a problem that needs to be urgently solved by technical personnel in this field.

[0063] Therefore, to solve the above problem, please refer to Figure 1 The embodiment of the present invention provides a method for generating marketing words, wherein the method comprises the steps of:

[0064] S100, collecting elements required for generating product marketing scripts as input data;

[0065] S200, based on the input data, generating corresponding specific marketing words as output data through a preset large language model;

[0066] S300, based on the input data and the output data, generating a corresponding marketing speech generation model by fine-tuning a preset pre-training model;

[0067] S400: Generate target marketing words required for the target product based on the marketing words generation model.

[0068] During specific implementation, in this embodiment, the elements required to generate the marketing language of the product are collected as input data, wherein the product refers to a single product. The method of the present invention is mainly aimed at insurance products, and of course, it can also be other types of single products.

[0069] Furthermore, in the marketing script generation method, the step S100 of collecting elements required for generating the marketing script of the product as input data includes:

[0070] Collect existing marketing scripts for several products, analyze and disassemble the existing marketing scripts, obtain the elements required to generate the marketing scripts for the corresponding products, and use the elements as input data; wherein the elements include: the theme, the key points of the theme's scripts, product information, and the marketing group.

[0071] Furthermore, in the marketing script generation method, the product information includes: product name, product highlights, and guarantee responsibilities.

[0072] In the specific implementation, in this embodiment, the elements required for generating the marketing words of the product are collected as input data, wherein the specific implementation process is as follows:

[0073] 1) Collect marketing words

[0074] Collect a certain number of existing marketing scripts that cover different marketing scenarios for insurance products;

[0075] 2) Analysis and disassembly

[0076] Conduct in-depth analysis of the existing marketing scripts collected to identify their structure, language style, persuasion techniques, etc., and disassemble each script to extract its core components;

[0077] 3) Extract elements

[0078] a. Theme: the specific scenario in which marketing tactics are applied;

[0079] This section mainly collects 10 topics including product introduction, stimulating demand, promoting sales, referrals, objections (such as too expensive, want to compare with other companies, etc.);

[0080] b. Key points of speech under each topic

[0081] Different topics require corresponding speech strategies, so it is necessary to extract the key points of each topic. This part uses the summarization capability of the large language model (LLM) to first extract the corresponding speech key points for each topic of existing popular products, then cluster the speech key points of the same topic, extract a small number of key points from each class, and finally manually select some representative key points as the key points of the topic (each topic only needs to confirm the key points of the topic once);

[0082] c. Product information

[0083] Collect product information of a certain number of insurance products. This part mainly uses the information extraction capability of the large language model (LLM). First, extract the marketing population from the product information of the insurance products (including product name, product highlights, coverage, liability details, coverage, service details, etc.), and then extract the main highlights and main responsibilities of the insurance products based on the marketing population.

[0084] d. Marketing crowd

[0085] In the marketing field, "target audience" refers to the specific consumer group that a company or brand hopes to attract and influence through marketing activities;

[0086] 4) Elements as input data

[0087] The extracted elements can be organized into formatted data, which can be used as input data for model learning or to generate new marketing words;

[0088] 5) Data annotation (if necessary):

[0089] If you need the model to understand specific industry terms or marketing strategies, you can annotate the data to enhance the model's recognition and application capabilities;

[0090] Through this process, the elements required for generating marketing scripts for insurance products can be obtained and used as input data.

[0091] Furthermore, the marketing speech generation method, wherein the step S200, based on the input data, generates corresponding specific marketing speech as output data through a preset large language model, includes:

[0092] Inputting the input data into a preset large language model to generate corresponding initial marketing words;

[0093] After data enhancement processing is performed on the initial marketing words, corresponding specific marketing words are obtained, and the specific marketing words are used as output data.

[0094] In the specific implementation, in this embodiment, based on the input data, a preset large language model is used to generate corresponding specific marketing words as output data, wherein the specific implementation process is as follows:

[0095] 1) Input the preset large language model

[0096] Input the sorted input data into a preset large language model, such as Qwen-72b, to generate the corresponding initial marketing words;

[0097] 2) Generate initial marketing script

[0098] Using the text generation capability of the large language model, a series of initial marketing scripts for specific scenarios and groups are generated in zero-shot based on the input elements.

[0099] 3) Data enhancement processing

[0100] Perform data enhancement on the generated initial marketing words to increase the diversity and adaptability of the words; data enhancement may include: enriching the vocabulary by replacing synonyms, changing the sentence structure to increase the diversity of expression, etc.

[0101] 4) Generate specific marketing words

[0102] According to the results of data enhancement processing, adjust and optimize the sales pitch to obtain the corresponding specific marketing sales pitch;

[0103] 5) Specific marketing words as output data

[0104] The specific marketing words after data enhancement and optimization are used as output data, which can be used as output data for model learning or further analysis and evaluation;

[0105] 6) Data annotation (if necessary):

[0106] If you need the model to understand specific industry terms or marketing strategies, you can annotate the data to enhance the model's recognition and application capabilities;

[0107] Through this process, the large language model can be effectively used to generate and optimize marketing words, obtain corresponding specific marketing words, and use the specific marketing words as output data.

[0108] Furthermore, the marketing speech generation method, wherein the step S300, based on the input data and the output data, generates a corresponding marketing speech generation model by fine-tuning a preset pre-trained model, comprises:

[0109] Based on the input data and the output data, the Qwen-14B model is fine-tuned by using the loRA method to generate a corresponding initial model;

[0110] The initial model is verified and tested, and when it is detected that the verification result and the test result meet the preset requirements, a corresponding marketing speech generation model is generated.

[0111] In the specific implementation, in this embodiment, based on the input data and the output data, the preset pre-training model is fine-tuned to generate a corresponding marketing speech generation model, wherein the specific implementation process is as follows:

[0112] 1) Data preparation

[0113] Collect and organize input data (such as product information, marketing population, etc.) and expected output data (specific marketing words) for fine-tuning;

[0114] 2) Preprocessing

[0115] Preprocess input and output data, including cleaning, formatting, and word segmentation, to meet the input requirements of the model;

[0116] 3) Load the pre-trained model

[0117] Load the Qwen-14B model, which is a large pre-trained language model with 1.4 billion parameters;

[0118] 4) Apply the LoRA method

[0119] Apply the LoRA (Low-Rank Adaptation) method to fine-tune the Qwen-14B model, which involves introducing low-rank matrices in certain layers of the model to adjust the model parameters;

[0120] 5) Fine-tuning training

[0121] Fine-tune the model using the prepared dataset; during the training process, only the parameters of the LoRA adapter are updated, while the original model parameters are kept unchanged or partially updated;

[0122] 6) Model Validation

[0123] Evaluate the performance of the fine-tuned model on the validation set to check whether the marketing words it generates meet expectations;

[0124] 7) Model testing

[0125] Further evaluate the fine-tuned model on the test set to ensure that it performs well on unseen data;

[0126] 8) Performance meets requirements

[0127] When the performance of the fine-tuned model in both the validation and testing phases meets the preset requirements, the fine-tuning is considered successful;

[0128] 9) Generate a marketing speech generation model

[0129] Save the fine-tuned model as a marketing script generation model for generating personalized targeted marketing scripts in actual applications;

[0130] Through this process, a marketing script generation model customized for specific marketing needs (insurance vertical field) can be created. This model can automatically generate effective marketing scripts based on product characteristics and marketing audiences, which can significantly improve the efficiency and personalization level of marketing activities.

[0131] Furthermore, in the marketing speech generation method, the step S400 of generating the target marketing speech required for the target product based on the marketing speech generation model includes:

[0132] Obtain product information corresponding to the target product given by the target user;

[0133] Based on the corresponding product information, determine the target theme and target marketing points of the target product, as well as the target marketing population;

[0134] The corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product.

[0135] In specific implementation, in this embodiment, based on the marketing speech generation model, the target marketing speech required for the target product is generated, wherein the specific implementation process is as follows:

[0136] 1) Get the product information corresponding to the target product

[0137] Obtain corresponding product information from the target product (target insurance product) given by the target user;

[0138] 2) Determine marketing goals

[0139] Determine the target theme of marketing activities based on the corresponding product information, such as product promotion, brand awareness enhancement, specific promotional activities, etc.;

[0140] 3) Key points for formulating target speech

[0141] Identify the key information points that need to be highlighted in marketing language, such as the product’s unique selling points, user benefits, calls to action, etc.

[0142] 4) Identify target marketing groups

[0143] Identify and analyze the characteristics of the target audience, including their age, gender, interests, consumption habits, etc.;

[0144] 5) Input information to the generative model

[0145] Provide the corresponding product information, target theme, target marketing key points, and target marketing population as input data to the marketing marketing generation model;

[0146] 6) Generate marketing words

[0147] Use the marketing script generation model to process the input data and generate targeted marketing scripts for target products;

[0148] Through this process, we can ensure that the target marketing language is not only consistent with the characteristics and market positioning of the target product, but also can resonate with the target marketing population and improve the effectiveness of marketing activities; using the marketing language generation model can greatly improve efficiency while maintaining the personalization and innovation of the language.

[0149] Furthermore, the marketing speech generation method, wherein the corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product, comprises:

[0150] The target marketing words are sorted out and returned to the target user.

[0151] In the specific implementation, in this embodiment, the target marketing words are sorted and returned to the target user, wherein the specific implementation process is as follows:

[0152] 1) Speech review

[0153] Before returning to the target users, conduct a final review of the generated targeted marketing scripts to ensure that there are no errors and that they are in line with the brand image and marketing strategy;

[0154] 2) Format adjustment

[0155] Adjust the format of your marketing campaign based on the scenario in which it will be used. For example, if it is for social media, it may require a more concise and eye-catching format; if it is for email marketing, it may require more detailed instructions and a clear call to action (CTA).

[0156] 3) Personalization

[0157] If necessary, targeted marketing pitches can be customized to ensure they meet the specific needs of different user groups or individuals;

[0158] 4) Organize and archive

[0159] Organize and archive the target marketing pitches into a format that is convenient for target users, such as documents, PPT slides, or online content;

[0160] 5) Provide usage guide

[0161] Provide guidance to target users on how to effectively use these targeted marketing techniques, including recommended application scenarios, best practices, and any considerations;

[0162] 6) Feedback Mechanism

[0163] Establish a feedback mechanism to allow target users to report their experience using the target marketing words and how effective the words are in actual application;

[0164] 7) Delivery method selection

[0165] Choose the appropriate delivery method based on the target user's preferences, such as email, online shared documents, or direct download links;

[0166] 8) User communication

[0167] Communicate with target users to ensure they understand how to receive and use these targeted marketing messages;

[0168] 9) Technical support

[0169] Provide necessary technical support to help target users solve any technical problems encountered in the process of using target marketing tactics;

[0170] Through this process, it can be ensured that target users receive high-quality targeted marketing scripts and can easily apply these scripts to their marketing activities, thereby improving marketing effectiveness and user satisfaction.

[0171] It can be seen from the above method embodiments that the marketing speech generation method provided by the present invention at least includes: collecting the elements required to generate the marketing speech of the product as input data; based on the input data, generating the corresponding specific marketing speech as output data through a preset large language model; based on the input data and the output data, fine-tuning the preset pre-trained model to generate a corresponding marketing speech generation model; based on the marketing speech generation model, generating the target marketing speech required for the target product; and so on. In this way, through the method of the present invention, a marketing speech generation model in a vertical field can be obtained through training to generate personalized target marketing speech required for the target product, thereby reducing the marketing operation cost of the target product.

[0172] It should be understood that, although the present application provides method operation steps as described in the embodiments or flowcharts, more or less operation steps may be included based on conventional or non-creative labor, and these operation steps are not necessarily performed in sequence according to the embodiment or flowchart. The order of steps listed in the embodiment or flowchart is only one way of executing the order of many steps, and does not represent the only execution order. It should be noted that there is not necessarily a certain order between the above steps. A person of ordinary skill in the art can understand from the description of the embodiment of the present invention that in different embodiments, the above steps may have different execution orders, that is, they may be executed in parallel, or they may be executed in exchange, etc. Moreover, at least a part of the steps in the embodiment or flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but may be executed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but may be executed in turn, alternately or synchronously with other steps or at least a part of the sub-steps or stages of other steps.

[0173] Based on the above method embodiment, please refer to Figure 2 Another embodiment of the present invention further provides a marketing speech generation system, wherein the system comprises:

[0174] A collection module 11, used to collect elements required for generating marketing scripts for products as input data;

[0175] A first generating module 12 is used to generate corresponding specific marketing words as output data based on the input data through a preset large language model;

[0176] A fine-tuning module 13, configured to generate a corresponding marketing speech generation model by fine-tuning a preset pre-training model based on the input data and the output data;

[0177] The second generating module 14 is used to generate target marketing words required for the target product based on the marketing words generating model.

[0178] In the specific implementation, in this embodiment, the elements required to generate the marketing language of the product are collected as input data, wherein the product refers to a single product. The system of the present invention is mainly aimed at insurance products, and of course, it can also be other types of single products.

[0179] Furthermore, in the marketing script generation system, the elements required for collecting and generating the marketing script of the product as input data include:

[0180] Collect existing marketing scripts for several products, analyze and disassemble the existing marketing scripts, obtain the elements required to generate the marketing scripts for the corresponding products, and use the elements as input data; wherein the elements include: the theme, the key points of the theme's scripts, product information, and the marketing group.

[0181] Furthermore, in the marketing script generation system, the product information includes: product name, product highlights, and guarantee responsibilities.

[0182] In the specific implementation, in this embodiment, the elements required for generating the marketing words of the product are collected as input data, wherein the specific implementation process is as follows:

[0183] 1) Collect marketing words

[0184] Collect a certain number of existing marketing scripts that cover different marketing scenarios for insurance products;

[0185] 2) Analysis and disassembly

[0186] Conduct in-depth analysis of the existing marketing scripts collected to identify their structure, language style, persuasion techniques, etc., and disassemble each script to extract its core components;

[0187] 3) Extract elements

[0188] a. Theme: the specific scenario in which marketing tactics are applied;

[0189] This section mainly collects 10 topics including product introduction, stimulating demand, promoting sales, referrals, objections (such as too expensive, want to compare with other companies, etc.);

[0190] b. Key points of speech under each topic

[0191] Different topics require corresponding speech strategies, so it is necessary to extract the key points of each topic. This part uses the summarization capability of the large language model (LLM) to first extract the corresponding speech key points for each topic of existing popular products, then cluster the speech key points of the same topic, extract a small number of key points from each class, and finally manually select some representative key points as the key points of the topic (each topic only needs to confirm the key points of the topic once);

[0192] c. Product information

[0193] Collect product information of a certain number of insurance products. This part mainly uses the information extraction capability of the large language model (LLM). First, extract the marketing population from the product information of the insurance products (including product name, product highlights, coverage, liability details, coverage, service details, etc.), and then extract the main highlights and main responsibilities of the insurance products based on the marketing population.

[0194] d. Marketing crowd

[0195] In the marketing field, "target audience" refers to the specific consumer group that a company or brand hopes to attract and influence through marketing activities;

[0196] 4) Elements as input data

[0197] The extracted elements can be organized into formatted data, which can be used as input data for model learning or to generate new marketing words;

[0198] 5) Data annotation (if necessary):

[0199] If you need the model to understand specific industry terms or marketing strategies, you can annotate the data to enhance the model's recognition and application capabilities;

[0200] Through this process, the elements required for generating marketing scripts for insurance products can be obtained and used as input data.

[0201] Furthermore, the marketing speech generation system, wherein the step of generating corresponding specific marketing speech as output data based on the input data through a preset large language model, includes:

[0202] Inputting the input data into a preset large language model to generate corresponding initial marketing words;

[0203] After data enhancement processing is performed on the initial marketing words, corresponding specific marketing words are obtained, and the specific marketing words are used as output data.

[0204] In the specific implementation, in this embodiment, based on the input data, a preset large language model is used to generate corresponding specific marketing words as output data, wherein the specific implementation process is as follows:

[0205] 1) Input the preset large language model

[0206] Input the sorted input data into a preset large language model, such as Qwen-72b, to generate the corresponding initial marketing words;

[0207] 2) Generate initial marketing script

[0208] Using the text generation capability of the large language model, a series of initial marketing scripts for specific scenarios and groups are generated in zero-shot based on the input elements.

[0209] 3) Data enhancement processing

[0210] Perform data enhancement on the generated initial marketing words to increase the diversity and adaptability of the words; data enhancement may include: enriching the vocabulary by replacing synonyms, changing the sentence structure to increase the diversity of expression, etc.

[0211] 4) Generate specific marketing words

[0212] According to the results of data enhancement processing, adjust and optimize the sales pitch to obtain the corresponding specific marketing sales pitch;

[0213] 5) Specific marketing words as output data

[0214] The specific marketing words after data enhancement and optimization are used as output data, which can be used as output data for model learning or further analysis and evaluation;

[0215] 6) Data annotation (if necessary):

[0216] If you need the model to understand specific industry terms or marketing strategies, you can annotate the data to enhance the model's recognition and application capabilities;

[0217] Through this process, the large language model can be effectively used to generate and optimize marketing words, obtain corresponding specific marketing words, and use the specific marketing words as output data.

[0218] Furthermore, the marketing speech generation system, wherein the corresponding marketing speech generation model is generated by fine-tuning a preset pre-trained model based on the input data and the output data, comprises:

[0219] Based on the input data and the output data, the Qwen-14B model is fine-tuned by using the loRA method to generate a corresponding initial model;

[0220] The initial model is verified and tested, and when it is detected that the verification result and the test result meet the preset requirements, a corresponding marketing speech generation model is generated.

[0221] In the specific implementation, in this embodiment, based on the input data and the output data, the preset pre-training model is fine-tuned to generate a corresponding marketing speech generation model, wherein the specific implementation process is as follows:

[0222] 1) Data preparation

[0223] Collect and organize input data (such as product information, marketing population, etc.) and expected output data (specific marketing words) for fine-tuning;

[0224] 2) Preprocessing

[0225] Preprocess input and output data, including cleaning, formatting, and word segmentation, to meet the input requirements of the model;

[0226] 3) Load the pre-trained model

[0227] Load the Qwen-14B model, which is a large pre-trained language model with 1.4 billion parameters;

[0228] 4) Apply the LoRA method

[0229] Apply the LoRA (Low-Rank Adaptation) method to fine-tune the Qwen-14B model, which involves introducing low-rank matrices in certain layers of the model to adjust the model parameters;

[0230] 5) Fine-tuning training

[0231] Fine-tune the model using the prepared dataset; during the training process, only the parameters of the LoRA adapter are updated, while the original model parameters are kept unchanged or partially updated;

[0232] 6) Model Validation

[0233] Evaluate the performance of the fine-tuned model on the validation set to check whether the marketing words it generates meet expectations;

[0234] 7) Model testing

[0235] Further evaluate the fine-tuned model on the test set to ensure that it performs well on unseen data;

[0236] 8) Performance meets requirements

[0237] When the performance of the fine-tuned model in both the validation and testing phases meets the preset requirements, the fine-tuning is considered successful;

[0238] 9) Generate a marketing speech generation model

[0239] Save the fine-tuned model as a marketing script generation model for generating personalized targeted marketing scripts in actual applications;

[0240] Through this process, a marketing script generation model customized for specific marketing needs (insurance vertical field) can be created. This model can automatically generate effective marketing scripts based on product characteristics and marketing audiences, which can significantly improve the efficiency and personalization level of marketing activities.

[0241] Furthermore, the marketing speech generation system, wherein the generation of target marketing speech required for the target product based on the marketing speech generation model includes:

[0242] Obtain product information corresponding to the target product given by the target user;

[0243] Based on the corresponding product information, determine the target theme and target marketing points of the target product, as well as the target marketing population;

[0244] The corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product.

[0245] In specific implementation, in this embodiment, based on the marketing speech generation model, the target marketing speech required for the target product is generated, wherein the specific implementation process is as follows:

[0246] 1) Get the product information corresponding to the target product

[0247] Obtain corresponding product information from the target product (target insurance product) given by the target user;

[0248] 2) Determine marketing goals

[0249] Determine the target theme of marketing activities based on the corresponding product information, such as product promotion, brand awareness enhancement, specific promotional activities, etc.;

[0250] 3) Key points for formulating target speech

[0251] Identify the key information points that need to be highlighted in marketing language, such as the product’s unique selling points, user benefits, calls to action, etc.

[0252] 4) Identify target marketing groups

[0253] Identify and analyze the characteristics of the target audience, including their age, gender, interests, consumption habits, etc.;

[0254] 5) Input information to the generative model

[0255] Provide the corresponding product information, target theme, target marketing key points, and target marketing population as input data to the marketing marketing generation model;

[0256] 6) Generate marketing words

[0257] Use the marketing script generation model to process the input data and generate targeted marketing scripts for target products;

[0258] Through this process, we can ensure that the target marketing language is not only consistent with the characteristics and market positioning of the target product, but also can resonate with the target marketing population and improve the effectiveness of marketing activities; using the marketing language generation model can greatly improve efficiency while maintaining the personalization and innovation of the language.

[0259] Furthermore, the marketing speech generation system, wherein the corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product, comprises:

[0260] The target marketing words are sorted out and returned to the target user.

[0261] In the specific implementation, in this embodiment, the target marketing words are sorted and returned to the target user, wherein the specific implementation process is as follows:

[0262] 1) Speech review

[0263] Before returning to the target users, conduct a final review of the generated targeted marketing scripts to ensure that there are no errors and that they are in line with the brand image and marketing strategy;

[0264] 2) Format adjustment

[0265] Adjust the format of your marketing campaign based on the scenario in which it will be used. For example, if it is for social media, it may require a more concise and eye-catching format; if it is for email marketing, it may require more detailed instructions and a clear call to action (CTA).

[0266] 3) Personalization

[0267] If necessary, targeted marketing pitches can be customized to ensure they meet the specific needs of different user groups or individuals;

[0268] 4) Organize and archive

[0269] Organize and archive the target marketing pitches into a format that is convenient for target users, such as documents, PPT slides, or online content;

[0270] 5) Provide usage guide

[0271] Provide guidance to target users on how to effectively use these targeted marketing techniques, including recommended application scenarios, best practices, and any considerations;

[0272] 6) Feedback Mechanism

[0273] Establish a feedback mechanism to allow target users to report their experience using the target marketing words and how effective the words are in actual application;

[0274] 7) Delivery method selection

[0275] Choose the appropriate delivery method based on the target user's preferences, such as email, online shared documents, or direct download links;

[0276] 8) User communication

[0277] Communicate with target users to ensure they understand how to receive and use these targeted marketing messages;

[0278] 9) Technical support

[0279] Provide necessary technical support to help target users solve any technical problems encountered in the process of using target marketing tactics;

[0280] Through this process, it can be ensured that target users receive high-quality targeted marketing scripts and can easily apply these scripts to their marketing activities, thereby improving marketing effectiveness and user satisfaction.

[0281] As can be seen from the above system embodiments, the marketing speech generation system provided by the present invention at least includes: collecting the elements required to generate the marketing speech of the product as input data; based on the input data, generating the corresponding specific marketing speech as output data through the preset large language model; based on the input data and the output data, fine-tuning the preset pre-trained model to generate the corresponding marketing speech generation model; based on the marketing speech generation model, generating the target marketing speech required for the target product; and so on. In this way, the system of the present invention can obtain a marketing speech generation model in a vertical field through training to generate the personalized target marketing speech required for the target product, thereby reducing the marketing operation cost of the target product.

[0282] Based on the above method embodiment, please refer to Figure 3 Another embodiment of the present invention further provides a computer device, wherein the computer device 10 includes:

[0283] memory 120 and one or more processors 110, Figure 3 A processor 110 is used as an example for description. The processor 110 and the memory 120 may be connected via a communication bus or other means. Figure 3 The example of connecting via a communication bus is taken as an example.

[0284] The processor 110 is used to complete various control logics of the computer device 10. It can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, a discrete gate or transistor logic, a discrete hardware component, or any combination of these components. In addition, the processor 110 can also be any traditional processor, microprocessor or state machine. The processor 110 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a DSP, or any other such configuration.

[0285] The memory 120 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as the computer program corresponding to the marketing speech generation method in the embodiment of the present invention. The processor 110 executes various functional applications and data processing of the computer device 10 by running the non-volatile software programs, instructions and units stored in the memory 120, that is, implements the marketing speech generation method in the above method embodiment.

[0286] The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store an application required for operating the device or at least one function; the data storage area may store data created according to the use of the computer device 10, etc. In addition, the memory 120 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 120 may optionally include a memory remotely arranged relative to the processor 110, and these remote memories may be connected to the computer device 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0287] One or more units are stored in the memory 120. When executed by one or more processors 110, the marketing speech generation method in any of the above method embodiments can be implemented. For example, the above described marketing speech generation method can be implemented. Figure 1 The method comprises steps S100 to S400.

[0288] Those skilled in the art will understand that Figure 3 The hardware structure diagram shown in the figure is only a schematic diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0289] Based on the above method embodiments, another embodiment of the present invention further provides a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the marketing speech generation method in any of the above method embodiments can be implemented, for example, the above described Figure 1 The method comprises steps S100 to S400.

[0290] As an example, the non-volatile storage medium can include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable ROM (EEPROM), or a flash memory. Volatile memory can include a random access memory (RAM) as an external cache memory. By way of illustration and not limitation, RAM can be obtained in many forms such as synchronous RAM (SRAM), dynamic RAM, (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synch link DRAM (SLDRAM), and direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable types of memory.

[0291] Based on the above method embodiments, another embodiment of the present invention further provides a computer program product, the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, the computer program includes program instructions, when the program instructions are executed by a processor, the marketing speech generation method in any of the above method embodiments can be implemented, for example, the above described Figure 1 The method comprises steps S100 to S400.

[0292] The above described embodiments are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.

[0293] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can exist in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0294] Conditional language such as "can," "can," "might," or "may," unless specifically stated otherwise or otherwise understood within the context as used, is generally intended to convey, among other things, that a particular embodiment can include (while other embodiments do not) a particular feature, element, and / or operation. Thus, such conditional language is also generally intended to imply that a feature, element, and / or operation is required for one or more embodiments anyway or that one or more embodiments must include logic for determining, with or without input or prompting, whether such features, elements, and / or operations are included or will be performed in any particular embodiment.

[0295] What has been described herein, in this specification and in the accompanying drawings includes examples that can provide a method, system, computer device, and non-volatile computer-readable storage medium for generating marketing speech. Of course, it is not possible to describe every conceivable combination of elements and / or methods for the purpose of describing the various features of the present disclosure, but it can be recognized that many other combinations and permutations of the disclosed features are possible. Therefore, it is obvious that various modifications can be made to the present disclosure without departing from the scope or spirit of the present disclosure, but all these various modifications should belong to the scope of protection of the claims attached to the present invention. In addition, or in an alternative, other embodiments of the present disclosure may be obvious from consideration of the present specification and the accompanying drawings and from the practice of the present disclosure as presented herein. It is intended that the examples proposed in the present specification and the accompanying drawings are considered to be illustrative and not restrictive in all respects. Although specific terms are used herein, they are used in a general and descriptive sense and are not used for limiting purposes.

Claims

1. A method for generating marketing words. It is characterized in that include: Collect the elements needed to generate product marketing pitches as input data; Based on the input data, a preset large language model is used to generate corresponding specific marketing words as output data; Based on the input data and the output data, a corresponding marketing speech generation model is generated by fine-tuning a preset pre-trained model; Based on the marketing script generation model, target marketing scripts required for the target product are generated.

2. The method for generating marketing words according to claim 1, It is characterized in that The elements required for collecting and generating the marketing scripts of the product as input data include: Collect existing marketing scripts for several products, analyze and disassemble the existing marketing scripts, obtain the elements required to generate the marketing scripts for the corresponding products, and use the elements as input data; wherein the elements include: the theme, the key points of the theme's scripts, product information, and the marketing group.

3. The method for generating marketing words according to claim 2, It is characterized in that Based on the input data, generating corresponding specific marketing words as output data through a preset large language model includes: Inputting the input data into a preset large language model to generate corresponding initial marketing words; After data enhancement processing is performed on the initial marketing words, corresponding specific marketing words are obtained, and the specific marketing words are used as output data.

4. The method for generating marketing words according to claim 3, It is characterized in that The method of generating a corresponding marketing speech generation model by fine-tuning a preset pre-trained model based on the input data and the output data includes: Based on the input data and the output data, the Qwen-14B model is fine-tuned by using the loRA method to generate a corresponding initial model; The initial model is verified and tested, and when it is detected that the verification result and the test result meet the preset requirements, a corresponding marketing speech generation model is generated.

5. The method for generating marketing words according to claim 4, It is characterized in that The step of generating target marketing words required for the target product based on the marketing words generation model includes: Obtain product information corresponding to the target product given by the target user; Based on the corresponding product information, determine the target theme and target marketing points of the target product, as well as the target marketing population; The corresponding product information, the target theme, the target speech key points, and the target marketing population are input into the marketing speech generation model to generate the target marketing speech required for the target product.

6. The method for generating marketing words according to claim 5, It is characterized in that After inputting the corresponding product information, the target theme, the target marketing key points, and the target marketing population into the marketing marketing generation model to generate the target marketing required by the target product, the method includes: The target marketing words are sorted out and returned to the target user.

7. The method for generating marketing words according to any one of claims 2 to 6, It is characterized in that The elements include: theme, key points of the theme, product information, and marketing groups, among which the product information includes: product name, product highlights, and guarantee responsibilities.

8. A marketing speech generation system, It is characterized in that include: A collection module is used to collect the elements required to generate the marketing script of the product as input data; A first generating module is used to generate corresponding specific marketing words as output data through a preset large language model based on the input data; A fine-tuning module, configured to generate a corresponding marketing speech generation model by fine-tuning a preset pre-training model based on the input data and the output data; The second generation module is used to generate target marketing words required for the target product based on the marketing words generation model.

9. A computer device, It is characterized in that The computer device comprises at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor. When the computer program is executed by the at least one processor, the marketing speech generation method as described in any one of claims 1 to 7 can be implemented.

10. A non-volatile computer-readable storage medium, It is characterized in that The non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by at least one processor, the marketing speech generation method as described in any one of claims 1 to 7 can be implemented.

Images